310 BULLETIN 82, UNITED STATES NATIONAL MUSEUM. 



creature with five arms like Pentametrocrinus and a centrodorsal bearing five cirri, 

 in cverj' way like the nodal of an Isocrinus. 



We should imagine that the increase in the number of the cirri took place by a 

 process similar to, though entirol}' independent of, the method of reduplication of 

 the arms, and that the first step was a pairing or twiiming of the primitive cirrus 

 elements, whereby 2 cirri, just alike, were produced in each radial area instead 

 of the original 1, exactly as the 10 arms of most comatulids arose from the orig- 

 inal 5. Each of the 10 arms in the various pairs is practically the exact duplicate 

 of its fellow, and the pairs are separated from the radial by the interpolation of 2 

 ossicles which are reduplicated repetitions of the 2 first ossicles in either arm, 

 which themselves are a pair of twins derived jihylogenetically from the first 2 

 ossicles of the primitive unpaired arm, this in turn being the resultant from 2 

 pairs of primitive ambulacral plates. 



In the case of the cirri the di%'ision of the originally single cirrus into two would 

 take place at the base, as in the case of the arms, but the base is entirely within the 

 centrodorsal, and usually within the free central cavity so that the cirri, instead 

 of appearing externaUy as a paired organ appear as two similar organs side by side, 

 usually slightty displaced by crowding. Further reduplication of the cirri might 

 have been carried on in either of two ways: (1 ) A more or less continuous budding 

 might take place, the original cirrus stem putting forth additional cirri as a tree puts 

 forth branches; or (2) the paired condition may be reduplicated, giving rise to cirri 

 in paired columns. 



By this reasoning we see how the body appendages, both the arms and the 

 cirri, reduce themselves each to a single simple linear series of essentially similar 

 segments ; that is, to a pair of such appendages to each half somite, comparable to 

 the paired somatic appendages of the crustaceans. No comatulid is highly special- 

 ized, and none are primitive, in all their characters, but each type is composed of 

 characters some of which are highly specialized whUe the remainder are primitive, 

 the characters changing their relative balance in each group, though a general 

 balance is observable every\^'here. In the comatulids as we know them, that is, 

 without regard to their phylogenetic history, the very large centrodorsal with ex- 

 ceedmgly numerous cirri is probably the most primitive type, as most nearly ap- 

 proaching the conditions found m the closely related pentaci inites, but this is always 

 associated with a high grade of specialization in other structures. Conversely, 

 the most primitive type of comatulid arm is invariably found with highly- specialized 

 cini and an enormouslj^ developed musculature. 



The relationship of the chief types of cirri to the larger systematic groups is 

 briefly shown m the followuig table: 



A. Short, stout and smooth cirri, with a small number of similar and suheciual 

 segments. 



B. Longer cirri with more numerous segments, of which (he distal are shorter 

 than the proximal and bear dorsal processes. 



C. Enormously elongated cirri, with the same structure as those grouped 

 under B. 



